Jack Wilkinson

Applied Medical Statistician. NIHR Doctoral Research Fellow. Statistical Editor for Cochrane Gynaecology and Fertility. IVF Researcher.

Recent Comments

Jun 29, 2017

Jack Wilkinson, Centre for Biostatistics,Institute of Population Health, Manchester Academic Health Science Centre (MAHSC), University of Manchester, Manchester, UK.

Sarah Lensen, Department of Obstetrics and Gynaecology, University of Auckland, New Zealand.

Nyboe Andersen and colleagues (2017) present a large randomised study comparing follitropin delta, with dosing based on AMH and body weight, against follitropin alfa, with dose selection by conventional means. Stated key findings of the study include the fact that more women achieved a target response of 8-14 oocytes (43.3% vs 38.4%) in the follitropin delta group, and that fewer women had poor (< 4 oocytes) or excessive (>= 15 or >= 20 oocytes) responses compared to the follitropin alfa group. However, on close inspection of the published study, it is not clear that follitropin delta, administered using this dose selection algorithm, has been shown to be superior to the standard dose of follitropin alfa in relation to response to stimulation. 

We note that calculations regarding number of oocytes obtained do not include women who were randomised but subsequently had their stimulation cycle cancelled for anticipated poor response (prior to hCG trigger). Indeed, there were more women with cycles cancelled for anticipated poor response in the follitropin delta group than the follitropin alfa group. The claim that poor response was lower in the follitropin delta arm then appears to be technically correct but potentially highly misleading. It would be preferable to include women with cancelled cycles in the numerator and denominator (they appeared in neither numerator nor denominator in the analysis presented by Nyboe Andersen and colleagues) by setting the number of oocytes recovered for women with cancelled stimulation cycles to be zero. This would rectify any bias resulting from differential cancellation rates between the study arms. It would also preserve the balance over confounding factors produced by randomisation, which is otherwise violated. When we include all patients in this way, the mean number of oocytes retrieved in the follitropin delta and follitropin alfa arms is 9.6 and 10.1, respectively, and the numbers achieving the target response are 275 (41%) and 247 (37%), which gives p=0.15 from a chi-squared test. 

A second concern is the fact that the authors present the rate of poor response in the low AMH group and the rate of hyper response in the high AMH group, but do not present the rate of poor response in the high AMH group or the rate of hyper response in the low AMH group. If we refer to hyper responses in low AMH patients and poor responses in high AMH patients as ‘inverted’, then we can calculate the number of inverted responses in each group. In fact, the number of inverted responses are greater in the follitropin delta group (5% vs 4% using 15 oocytes as the threshold for hyper response, or 4% vs 2% using 20 oocytes):

Number Inverted <4 or >=15 is 34/640 (5%) in experimental arm and 25/643 (4%) in the control arm, p=0.233 from Fisher's exact test.        
Number Inverted <4 or >=20 is 23/640 (4%) in experimental arm and 11/643 (2%) in the control arm, p=0.038 from       Fisher’s exact test.

These calculations are conducted in patients achieving hCG trigger, as it does not appear to be possible to identify whether the women with cycles cancelled for poor response were in the high or low AMH stratum from the information presented. The number of hyper responses in the low AMH group is of particular interest, since this represents increased risk to the treated woman and any resulting offspring. Unfortunately, this information is obfuscated by the presentation in the manuscript. We also note that the decision to report the rates of hypo and hyper response only in these subgroups appears to be a departure from the clinicaltrials.gov record for the trial (https://clinicaltrials.gov/ct2/show/NCT01956110).

Given that the claim of the authors in relation to the effectiveness of treatment was one of non-inferiority, the matter of the safety of the individualised dosing regimen compared to standard dosing would appear to be of paramount importance. On the basis of the considerations outlined above, it is unclear that an advantage of the individualised follitropin delta regimen in relation to achieving target ovarian responses has been demonstrated. Moreover, the data leave open the possibility that the individualised regimen may increase excessive responses in low AMH patients. We would like to invite the authors to clarify this point. 

A version of this comment has been posted on Pubmed Commons.

Nyboe Andersen, A., et al. (2017). "Individualized versus conventional ovarian stimulation for in vitro fertilization: a multicenter, randomized, controlled, assessor-blinded, phase 3 noninferiority trial." Fertil Steril 107(2): 387-396 e384.

Conflict of interest statement

JW is funded by a Doctoral Research Fellowship from the National Institute for Health Research (DRF-2014-07-050). The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health. JW also declares that publishing research is beneficial to his career. JW is a statistical editor for the Cochrane Gynaecology and Fertility Group, although the views expressed here are not necessarily those of the group.